3 ms·
I think this class of problems can be protected against. It's become clear that the first and most important and most valuable agent, or team of agents, to bui
by observationist 11mo ago
I think this class of problems can be protected against.
It's become clear that the first and most important and most valuable agent, or team of agents, to build is the one that responsibly and diligently lays out the opsec framework for whatever other system you're trying to automate.
A meta-security AI framework, cursor for opsec, would be the best, most valuable general purpose AI tool any company could build, imo. Everything from journalism to law to coding would immediately benefit, and it'd provide invaluable data for post training, reducing the overall problematic behaviors in the underlying models.
Move fast and break things is a lot more valuable if you have a red team mechanism that scales with the product. Who knows how many facepalm level failures like this are out there?
- croes 11mo ago> I think this class of problems can be protected against. Of course, it’s called proper software development
- jeffbee 11mo agoThe techniques for non-disclosure of confidential materials processed by multi-tenant services are obvious, well-known, and practiced by very few.
- venturecruelty 11mo agoAnd jail time for executives who are responsible for data leaks.
- marginalx 11mo agoAre you saying executives cannot make mistakes ever (ask because you didn't qualify your statement)?
- venturecruelty 11mo agoI'm saying that if executives get praise and bonuses for when good things happen, they should also have negative consequences when bad things happen. Litigate that further how you wish.
- dghlsakjg 11mo agoThe key word in that is "responsible". The legal world has plenty of ways for determining if you are legally responsible for the outcome of an event. Right now the standard is civil punishments for provable negligence. It sounds like GP is proposing a framework where we tighten up the definition of negligence, and add criminal penalties in addition to civil ones.
- pbhjpbhj 11mo agoAre you saying the OP was just a single error, effectively an executives typo.
- dghlsakjg 11mo agoIn this case, AI was a red herring. This was just plain terrible web security.